Showing papers on "Object detection published in 1983"

Application of Three-Dimensional Filtering to Moving Target Detection

Abstract: The standard approach to the detection of a stationary target immersed within an optically observed scene is to use integration to separate the target energy from the background clutter. When the target is nonstationary and moves with fixed velocity relative to the clutter, the procedure for integrating the target signal is no longer obvious. In this paper it is shown that the problem of tracking a target having a fixed velocity can be cast into a general framework of three-dimensional filter theory. From this point of view, the target detection problem reduces to the problem of finding optimal three-dimensional filters in the three-dimensional transform domain and processing the observed scene via this filtering. The design of these filters is presented, taking into account the target, clutter, and optical detection models. Performance is computed for a basic clutter model, showing the effective increase in detectability as a function of the target velocity. The three-dimensional transform approach is readily compatible with VLSI array processing technology.

Detection Performance in Clutter with Variable Resolution

Abstract: Experiments were conducted to determine the influence of background clutter on target detection criteria. The experiment consisted of placing observers in front of displayed images on a TV monitor. Observer ability to detect military targets embedded in simulated natural and manmade background clutter was measured when there was unlimited viewing time. Results were described in terms of detection probability versus target resolution for various signal to clutter ratios (SCR). The experiments were preceded by a search for a meaningful clutter definition. The selected definition was a statistical measure computed by averaging the standard deviation of contiguous scene cells over the whole scene. The cell size was comparable to the target size. Observer test results confirmed the expectation that the resolution required for a given detection probability was a continuum function of the clutter level. At the lower SCRs the resolution required for a high probability of detection was near 6 line pairs per target (LP/TGT), while at the higher SCRs it was found that a resolution of less than 0.25 LP/TGT would yield a high probability of detection. These results are expected to aid in target acquisition performance modeling and to lead to improved specifications for imaging automatic target screeners.

Viewer Independent Shape Recognition

Abstract: An important problem in vision is to detect the presence of a known rigid 3-D object. The general 3-D object recognition task can be thought of as building a description of the object that must have at least two parts: 1) the internal description of the object itself (with respect to an object-centered frame); and 2) the transformation of the object-centered frame to the viewer-centered (image) frame. The reason for this decomposition is parsimony: different views of the object should have minimal impact on its description. This is achieved by factoring the object's description into two sets of parameters, one which is view-independent (the object-centered component) and one which is view-varying (the viewing transformation). Often a description of the object is known beforehand and the task reduces to finding the objectframe to viewer-frame transformation. This paper describes a method for handling this case: a known object is detected by finding changes in orientation, translation, and scale of the object from its canonical description. The method is a Hough technique and has the characteristic insensitivity to occlusion and noise.

Reconstruction from projections based on detection and estimation of objects

Abstract: This paper considers the problem of observing a 2D function via its 1D projections (Radon transform); it presents a framework for detecting, locating and describing objects contained within a 2D cross-section by using noisy measurements of the Radon transform directly, rather than post-processing a reconstructed image. This framework offers the potential for significant improvements in applications where (1) attempts to perform an initial inversion with insufficient measurement data result in severely degraded reconstructions, and (2) the ultimate goal of the process is to obtain several specific pieces of information about the cross-section. To illustrate this perspective, we focus our attention on the problem of obtaining maximum-likelihood (ML) estimates of the parameters characterizing a single random object situated within a deterministic background medium, and we investigate the performance, robustness, and computational structure of the ML estimation procedure.

Variations In Task And The Ideal Observer

Abstract: In most previous studies involving the ideal observer, the task considered has been that of simple detection where it is assumed that there is complete a priori knowledge of the background and of the possible object's shape, amplitude, and position. It is shown that redefining the detection task to include the possibility of an unknown, slowly varying background reduces the importance of the low-frequency components in the image for the ideal observer. More complicated tasks than object detection are also considered, such as determination of an object's position and width and the resolution of two objects. These higher-order tasks further enhance the importance of the high-frequency information content of the image.

Fast Contour Detection Algorithm for High Precision Quantitative CT

Abstract: An algorithm for the automatic contour detection of objects in CT images is presented. It requires little memory space, allows easy incorporation of different types of local digital filters, and is fast. Typical computation time with a minicomputer is less than one second. An implementation of the algorithm for the detection of bone contours from low dose CT images is given as an example.

Novel Technologies for Automatic Focusing and White Balancing of Solid State Color Video Camera

Abstract: The consumer video market is rapidly spreading. In response, development of color video camera is strongly accelerated in these several years.

Object detection by two-dimensional linear prediction

Abstract: This paper addresses the problem of detecting small areas of textured images which differ from their immediate surroundings. A significance test is described which adapts itself to the generally changing background statistics so that a constant false alarm rate is maintained. A detection algorithm is derived from the fact that this significance test can be expressed in terms of the error residuals of an adaptive two-dimensional linear predictor whose coefficients are estimated from the background. The algorithm has been successfully demonstrated with both synthetic and real-world images.

A Holographic Approach for Motion Measurement

Abstract: Absfruct-In the areas of target tracking and biomedical tissue characterization, it is often important to detect or estimate the motion of an object. Slow and irregular multidimensional motion is difficult to determine by conventional approaches such as Doppler techniques. The presence of noise frequently makes space-domain analysis unrealizable. A frequency-domain approach is presented for resolving the trajectory of a moving object. The main body of this approach is the formulation of the optimization process and the optimal estimation of the displacement vector.

Holographic object detection

Abstract: PURPOSE:To conduct object detection with high resolution in the bearing and distance without depending on the detected distance by processing a hologram data obtained from received reflection waves, using specified test functions. CONSTITUTION:Received wave reflection data from an object by means of a transmission wave with many frequency components from each measurement point, as an anthena 2 mounted on an anthena transferring device 9 and a received wave transmitting device 3 move, is supplied to an accumulation device 6 through an A/D converter 4, a specified digital filter 5 with the measurement point location data through a position measurement device 7 and an A/D converter 8. The hologram data based on those received wave data is processed by an integrator 11 to which the test function from a function generating device 10 by which an image amplitude function becomes maximum that is defined by taking as variables the space coordinates between the test function of a function of two variables representing a position on a plane and a time variable and a received wave shape function. In this way the image regeneration is made by the image regeneration argonism based on pulse compression and aperture composition by clear formulation of the wave shapes reflected from the object.

The Relax Image Relaxation System: Description and Evaluation

Abstract: : RELAX is a system of routines that modifies the probabilities associated with labels attached to the elements of a two-dimensional array. These modifications reflect the compatibility of each element's labels with those of its neighbors. The initial probability assignments are usually derived from local property values in the neighborhood of each pixel. The final assignments may be used for object detection or segmentation, or may be mapped back to image intensities to achieve noise suppression, enhancement, or segmentation. The relaxation package was contributed to the ARPA/DMA Image Understanding Testbed at SRI by the University of Maryland. This report summarizes applications for which RELAX is suited the history and nature of the algorithm, details of the Testbed implementation, the manner in which RELAX is invoked and controlled the type of results that can be expected and suggestions for further development. The document includes both a user's guide to the RELAX system and an evaluation of the algorithm.

Signal detection performance of lattice processors

Abstract: Frequency domain techniques for pulsed Doppler radar data for target detection has received significant attention in the last few Years, and techniques employing FFT and Maximum Entropy methods (MEM) are much in favour. Some recent work [1,2] directed at comparing FFT techniques with MEM seem to indicate anomalies in the use of high resolution spectral estimation for target detection. Most of these techniques employ a blind decomposition of phase detected radar returns into frequency components, and make assessments based on computer simulations. This paper presents new results in the use of lattice processors for detecting signals in noise. A new approach to processing Doppler returns is proposed which involves the application of thresholds to the reflection coefficients in the lattice processor. A study of the underlying probability distribution of the reflection coefficients is included, which allows an optimal choice of threshold setting for a given false alarm rate, and probability of detection.